Method and System for Network Traffic Matrix Analysis

1. A method of calculating at a network device, data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of one of the intermediate nodes, the method comprising: obtaining local data traffic measurements at each of said elbows, wherein said local data traffic measurements comprise data traffic flowing through the elbows based on data traffic arriving at the intermediate node via said ingress interface and leaving the intermediate node via said egress interface; and performing at a processor at the network device, said data traffic flow calculations utilizing said local data traffic measurements; wherein obtaining said local data traffic measurements comprises classifying and measuring said data traffic based on how said data traffic passes across the intermediate node, said measuring comprising measuring a proportion of said data traffic routed over each of the elbows.

2. The method of claim 1 wherein utilizing said local data traffic measurements comprises generating data traffic matrix information.

3. The method of claim 2 wherein generating data traffic matrix information comprises utilizing data traffic matrix inference.

4. The method of claim 2 wherein generating data traffic matrix information comprises utilizing data traffic matrix estimation.

5. The method of claim 2 wherein said data traffic matrix information comprises data for a subset of flows traversing the network.

6. The method of claim 2 further comprising generating end-to-end data traffic flow estimates.

7. The method of claim 6 wherein generating said end-to-end data traffic flow estimates comprises utilizing a gravity estimation function.

8. The method of claim 6 wherein generating said end-to-end data traffic flow estimates comprises utilizing a path load feedback estimation function.

9. The method of claim 2 wherein generating data traffic matrix information comprises utilizing topology and routing data.

10. The method of claim 9 wherein utilizing topology and routing data comprises determining what proportion of data traffic flow passes over said elbows.

11. The method of claim 2 wherein the data traffic matrix information comprises data for aggregate flows comprising a plurality of individual flows which traverse the network.

12. The method of claim 11 wherein said data for aggregate flows comprises estimates of said aggregate flows.

13. The method of claim 11 wherein said data for aggregate flows comprises bounds on said aggregate flows.

14. The method of claim 13 further comprising performing resiliency checks in the network.

15. The method of claim 11 further comprising generating aggregate flow estimates and wherein said aggregate flow estimates are based on said local data traffic measurements and data traffic flow estimation.

16. The method of claim 15 wherein generating said aggregate flow estimates comprises utilizing a gravity estimation function.

17. The method of claim 15 wherein generating said aggregate flow estimates comprises utilizing a path load feedback estimation function.

18. The method of claim 1 wherein utilizing said local data traffic measurements comprises generating a set of constraints.

19. The method of claim 18 further comprising utilizing linear programming to solve said set of constraints.

20. The method of claim 18 wherein generating a set of constraints comprises generating a constraint equation according to: ∑ u , v ∈ V ′ × V ′ ⁢ c u , v , i , j ⁢ x u , v = d i , j wherein: V′×V′: set of all ordered pairs of edge nodes or set of all flows; i,j: links connected to the intermediate node; c u,v,i,j : proportion of flow from source node u to destination node v routed over elbow i,j; d i,j : observed bandwidth of data traffic crossing elbow i,j; and x u,v : variable representing the bandwidth of flow from source node u to destination node v.

21. The method of claim 18 further comprising generating a second set of constraints based on the sum of data traffic entering and exiting the network at an edge node.

22. The method of claim 18 further comprising generating a second set of constraints based on data traffic flow through links connecting said nodes.

23. The method of claim 18 further comprising utilizing at least one known flow in the system to further refine said set of constraints.

24. A method of calculating at a network device, data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of the intermediate node, the nodes being connected to one another by links, the method comprising: obtaining local data traffic measurements including proportion of data traffic flow from one of said source nodes to one of said destination nodes which is routed over each of said elbows, and observed bandwidth of data traffic crossing each of said elbows; determining local estimates of flow for each of said elbows using said local data traffic measurements; and calculating at a processor at the network device, end-to-end data traffic flow estimates based on said local flow estimates.

25. The method of claim 24 wherein calculating end-to-end data traffic flow estimates comprises utilizing an optimization function.

26. The method of claim 25 wherein utilizing an optimization function comprises applying a weight to each flow in a set of network flows.

27. The method of claim 25 further comprising solving a linear programming problem utilizing said optimization function.

28. The method of claim 25 wherein said optimization function is based on the difference between the observed bandwidth of data traffic crossing said elbow divided by the number of flows on said elbow and the proportion of data traffic flow on said elbow multiplied by a variable representing the bandwidth of end-to-end flow over a path from one of said source nodes to one of said destination nodes.

29. The method of claim 25 wherein said optimization function is based on said proportion of data traffic flow on each of said links and said observed bandwidth of data traffic crossing each of said links.

30. The method of claim 25 wherein said optimization function is based on said proportion of data traffic flow on each of said elbows and said observed bandwidth of data traffic crossing each of said elbows.

31. The method of claim 25 further comprising utilizing a set of constraints from a traffic flow model along with said optimization function to generate data traffic matrix estimates.

32. The method of claim 31 wherein the traffic flow model is a link-based traffic flow model.

33. The method of claim 31 wherein the traffic flow model is an elbow-based traffic flow model.

34. The method of claim 25 wherein utilizing said optimization function comprises utilizing a variable o u,v,i for each flow and link and constraint equations according to: o u , v , i ≥ b i  F i  - a u , v , i × x u , v ; and o u , v , i ≥ a u , v , i × x u , v - b i  F i  wherein: F i : set of flows on link i; a u,v,i : proportion of flow from node u to node v which is routed over link i; b i : observed bandwidth of data traffic crossing link i; and x u,v : variable representing bandwidth of flow from node u to node v.

35. The method of claim 34 further comprising solving a linear programming problem with said optimization function and wherein said optimization function is defined as: minimize ⁢ ( ∑ i ∈ I ⁢ ∑ ( u , v ) ∈ F i ⁢ o u , v , i ) .

36. The method of claim 25 wherein calculating data traffic flow comprises utilizing a variable o u,v,i,j and constraint equations according to: o u , v , i , j ≥ b i , j  F i , j  - a u , v , i , j × x u , v ; and o u , v , i , j ≥ a u , v , i , j × x u , v - b i , j  F i , j  wherein: F (i,j) : set of flows which traverse elbow (i,j); a u,v,i,j : proportion of flow from node u to node v which is routed over elbow (i,j); b (i,j) : observed bandwidth of data traffic crossing elbow (i,j); and x u,v : variable representing bandwidth of flow from node u to node v.

37. The method of claim 36 further comprising solving a linear programming problem with said optimization function and wherein said optimization function is defined as: minimize ⁢ ( ∑ i ∈ I ⁢ ∑ ( u , v ) ∈ F i ⁢ o u , v , i + ∑ ( i , j ) ∈ E ⁢ ∑ ( u , v ) ∈ F i , j ⁢ o u , v , i , j ) .

38. A method of calculating at a network device, data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of the intermediate node, the nodes being connected to one another by links, the method comprising: obtaining local data traffic measurements including observed bandwidth of data traffic crossing each of said links or said elbows; and determining local estimates for at least a portion of data traffic flows in the network based on said obtained local data traffic measurements, wherein all flows which contribute to each of said local estimates are estimated to be equal to one another; calculating end-to-end data traffic flow estimates utilizing an optimization function at a processor at the network device; and utilizing a set of constraints from a traffic flow model along with said function to generate data traffic matrix estimates.

39. The method of claim 38 further comprising solving a linear programming problem with said optimization function.

40. A non-transitory computer readable storage medium encoded with a computer program containing computer executable codes for calculating data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of one of the intermediate nodes, the computer program comprising: code that obtains local data traffic measurements at each of said elbows, wherein said local data traffic measurements comprise data traffic flowing through the elbows based on data traffic arriving at the intermediate node via said ingress interface and leaving the intermediate node via said egress interface; and code that utilizes said local data traffic measurements in said data traffic flow calculations; wherein code that obtains said local data traffic measurements comprises code that classifies and measures said data traffic based on how said data traffic passes across the intermediate node, said code that measures comprising code that measures a proportion of said data traffic routed over each of the elbows.

41. The computer-readable storage medium of claim 40 wherein code that utilizes said local data traffic measurements comprises code that generates data traffic matrix information.

42. A non-transitory computer readable storage medium encoded with a computer program containing computer executable codes for calculating end-to-end data traffic flow estimates in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of the intermediate node, the nodes being connected to one another by links, the computer program comprising: code that obtains local data traffic measurements including proportion of data traffic flow from one of said source nodes to one of said destination nodes which is routed over each of said elbows and observed bandwidth of data traffic crossing each of said elbows; code that determines local estimates of flow for each of said elbows using said local data traffic measurements; code that calculates end-to-end data traffic flow estimates based on said local flow estimates; and a computer-readable medium that stores the codes.

43. An apparatus for calculating at a network device, data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of the intermediate node, the nodes being connected to one another by links, the apparatus comprising: a processor for: obtaining local data traffic measurements including proportion of data traffic flow from one of said source nodes to one of said destination nodes which is routed over each of said links or said elbows, and observed bandwidth of data traffic crossing each of said links or said elbows; determining local estimates of flow for each of said links or said elbows using said local data traffic measurements; and calculating end-to-end data traffic flow estimates based on said local flow estimates; and memory for storing said local data traffic measurements; wherein calculating end-to-end data traffic flow estimates comprises utilizing an optimization function and wherein said optimization function is based on the difference between the observed bandwidth of data traffic crossing said link or said elbow divided by the number of flows on said link or said elbow and the proportion of data traffic flow on said link or said elbow multiplied by a variable representing the bandwidth of end-to-end flow over a path from one of said source nodes to one of said destination nodes.

44. An apparatus for calculating at a network device, data traffic flow in a communications network comprising a plurality of source nodes, a plurality of destination nodes, and a plurality of intermediate nodes, each of said plurality of intermediate nodes including at least one elbow comprising one ingress interface and one egress interface of the intermediate node, the apparatus comprising: a processor for: obtaining local data traffic measurements at each of said elbows, wherein said local data traffic measurements comprise data traffic flowing through the elbows based on data traffic arriving at the intermediate node via said ingress interface and leaving the intermediate node via said egress interface; performing said data traffic flow calculations utilizing said local data traffic measurements; and generating end-to-end data traffic flow estimates, wherein generating said end-to-end data traffic flow estimates comprises utilizing a path load feedback estimation function; and memory for storing said local data traffic measurements.